Journal of Consulting and Clinical Psychology Copyright 1996 by the American Psychological Association, Inc.

1996. Vol. 64, No. 6. 1276-1284 0022-006X/96/$3.00

SPECIAL FEATURE

Characteristics of Effective Therapists: Further Analyses of Data

From the National Institute of Mental Health Treatment
of Depression Collaborative Research Program
Sidney J. Blatt and Charles A. Sanislow III David C. Zuroff
Yale University McGill University

Paul A. Pilkonis
University of Pittsburgh

Analysesof the data of the National Institute of Mental Health-sponsored Treatment of Depression
Collaborative Research Program have primarily examined the effects of types of treatment and pa-
tient characteristics on outcome, but scant attention has been directed toward evaluating the contri-
butions of the therapist. With an aggregate of residualizedtherapeutic changescores of the 5 primary
outcome measures for each patient at termination as an overallmeasure of improvement, an average
therapeutic effectivenessmeasure was derived for each of the 28 therapists based on the outcome of
the patients they saw in active treatment. The distribution of the therapists was divided into thirds,
and comparisons indicate that more effective therapists are more psychological minded, eschew
biological interventions (i.e., medication and electroconvulsivetherapy) in their ordinary clinical
practice, and expect outpatient treatment of depression to take longerthan did moderately and less
effectivetherapists.

The National Institute of Mental Health (NIMH)-sponsored
Treatment of Depression Collaborative Research Program
(TDCRP), a comprehensive, well-designed, and carefully con-
ducted multisite, randomized clinical trial, evaluated several
forms of brief ( 16-20 session) outpatient treatment for depres-
sion. Prior analyses of these data compared the relative efficacy
of four treatment conditions and the effects of patient charac-
teristics on outcome, but little attention has been directed to
evaluating the therapists and their contributions to the thera-
peutic process.
Therapists' contribution to treatment outcome has long been
of concern in psychotherapy research (e.g., Beutler, Crago, &
Arizmendi, 1986; Gurman & Razin, 1977). However, as noted
Sidney J. Blatt, Departments of Psychiatry and Psychology,Yale Uni-
versity; Charles A. Sanislow III, Department of Psychiatry, School of
Medicine, YaleUniversity;David C. Zuroff, Department of Psychology,
McGill University, Montreal, Quebec, Canada; and Paul A. Pilkonis,
Department of Psychiatry, Universityof Pittsburgh.
We are indebted to the National Institute of Mental Health (NIMH)
and the members of the Treatment of Depression Collaborative Re-
search Program (TDCRP) research team for making availablethe em-
pirical data of the TDCRE The analyses presented in this article were
conducted without support from or the endorsement of the NIMH.
Correspondence concerning this article should be addressed to Sid-
ney J. Blatt, Department of Psychiatry, School of Medicine, Yale Uni-
versity, 25 Park Street, GEB 613, New Haven, Connecticut 06519. Elec-
tronic mail may be sent via Internet to Blatts@PANTHEON.Yale.edu.
by Luborsky and his colleagues (Luborsky & Auerbach, 1985;
Luborsky et al., 1986; Luborsky, Diguer, McLellan, & Woody,
1995 ) and by others (e.g., Frank, 1959; Lambert, 1989; Najav-
its & Strupp, 1994) the therapist is an often neglected and
poorly understood variable (Beutler, Machado, & Neufeldt,
1994). Although some therapists consistently achieve better re-
sults than others (e.g., Lambert, 1989; Lafferty, Beutler, &
Crago, 1989; Luborsky, Woody, McLeUan, O'Brien, & Rosen-
zweig, 1982; Luborsky et al., 1986; Orlinsky & Howard, 1980),
few characteristic of more effective therapists have been identi-
fied. Consistent with earlier reviews (Parloff, Waskow, & Wolfe,
1978 ), Beutler et al. (1994) found that only a few qualities have
even a modest effect on outcome: cognitive level (Holloway &
Wampold, 1986), capacity to establish a therapeutic alliance
(e.g., Horvath & Symonds, 1991 ), a background in short-term
therapy (e.g., Lyons & Woods, 1991; Miller & Berman, 1983),
professional background (e.g., Smith, Glass, & Miller, 1980;
Stein & Lambert, 1984), and the lack of directiveness in treat-
ment (Svartberg & Stiles, 1991 ). On the basis of peer ratings,
Luborsky, McLellan, Woody, O'Brien, and Auerbach (1985)
found that therapists' psychological health and skill, as well as
interest in helping patients, correlated positively with their pa-
tients' improvement. Possibly because of the small sample size,
however, these relationships did not reach statistical signifi-
cance. The degree to which the therapist adhered to the treat-
ment manual and the degree to which the patient, in the third
treatment session, reported experiencing the therapist as help-
ful was also correlated with treatment outcome (see also Blatt,

1276
 CHARACTERISTICS OF EFFECTIVE THERAPISTS
Zuroff, Quinlan, & Pilkonis, 1996). Despite the limited num-
ber of therapists' qualities that can be identified as influencing
treatment outcome (Beutler et al., 1994), Luborsky et al.
(1986, pp. 509-510) still concluded that the "frequency and
size of the therapists' effects generally overshadowed any differ-
ences between different forms of t r e a t m e n t . . . ?' and that
therapist effectiveness should be evaluated in all psychotherapy
outcome studies.
Rather than attempting to relate qualities of therapists to out-
come, Luborsky et al. (1995) suggested that a more productive
approach might be to compare relatively effective and in-
effective therapists. Several such studies have been conducted
(e.g., Crits-Christoph, Baranackie, Kurcias, & Beck, 1991;
Crits-Christoph & Mintz, 1991; Luborsky et al., 1985, 1986,
1995; McLellan, Woody, Luborsky, & Goehl, 1988; Najavits &
Strupp, 1994; Orlinsky & Howard, 1980; Ricks, 1974) with
some success. In the present study, estimates of therapeutic
efficacy were obtained for the 28 therapists in the TDCRP and
attempts were made to identify characteristics of the more
effective therapists.
The TDCRP, a collaborative clinical trial, compared three
brief outpatient treatments for depression. In a 3 (Research
Site) × 4 (Treatment Condition ) factorial design, 239 randomly
assigned patients entered one of four treatment conditions: cog-
nitive-behavioral therapy (CBT), interpersonal therapy (IPT),
imipramine plus clinical management (IMI-CM) as a standard
reference, and pill placebo plus clinical management (PLA-
CM ) as a double-blind control. Clinical management (CM) was
included as part of the two medication conditions to monitor
possible side effects of the medication and to provide general
support and encouragement as a minimal therapeutic condition
to deal with ethical concerns about treating severely depressed
patients with placebo (Elkin, 1994). CM included "clinically
indicated and appropriate supportive psychotherapeutic mea-
sures and i n t e r v e n t i o n s . . , interventions related to specific or-
ganized systems of psychotherapy [ w e r e ] . . . not permitted"
(Fawcett, Epstein, Fiester, Elkin, & Autry, 1987, p. 320).
Therapists in the four treatment conditions at each research
site were experienced MD- and PhD-level clinicians (10 each
providing IPT and pharmacotherapy, and 8 providing CBT),
with an average of more than 11 years of experience. All thera-
pists received training in the treatment they provided in the re-
search protocol, and only therapists who met competency cri-
teria participated in the study. Tapes of sessions were reviewed
to assure adherence to treatment protocols, and therapists re-
ceived consultation during the study ( Elkin, 1994).
Patients were nonbipolar, nonpsychotic outpatients who met
research diagnostic criteria for major depressive disorder
(Spitzer, Endicott, & Robins, 1978) and who scored 14 or
higher on an augmented, 20-item version of the 17-item Ham-
ilton Rating Scale for Depression (HRSD; Hamilton, 1960,
1967). Measurement of treatment outcome included interview
and self-report measures of depression; (the HRSD and the
Beck Depression Inventory [ BDI ], respectively), interview and
self-report measures of general clinical functioning (Global As-
sessment Scale [GAS] and Hopkins Symptom Checklist
[ HSCL-90 ], respectively), and an interview measure of social
adjustment (Social Adjustment Scale; SAS; Weissman & Pay-
kel, 1974).
1277
Prior analyses of the TDCRP data indicate only modest
differences in therapeutic outcome among the three brief treat-
ments for depression; IPT and IMI-CM were somewhat more
effective than CBT but primarily with more seriously depressed
patients (Elkin et al., 1989). Consistent with prior research
(e.g., Crits-Christoph, 1992; Crits-Christoph & Mintz, 1991;
Luborsky et al., 1986; Miller & Berman, 1983; Smith et al.,
1980; Stiles, Shapiro, & Elliott, 1986), however, outcome in the
TDCRP appears to be more related to differences among pa-
tients and therapists than types of treatment. Several pretreat-
merit characteristics of patients predicted outcome in specific
treatment conditions (Sotsky et al., 1991) and pretreatment
perfectionism, as assessed by one of two factors of the Dysfunc-
tional Attitudes Scale (DAS; Weissman & Beck, 1978), had a
significant negative effect on outcome (Blatt, Quinlan, Pilkonis,
& Shea, 1995). In addition, the degree to which patients per-
ceived their therapist at the end of the second treatment hour as
empathic, caring, open, and sincere, as assessed on the Barrett-
Lennard Relationship Inventory (B-L RI; Barrett-Lennard,
1962, 1985), significantly predicted the patient's outcome at
termination and reduced significantly the negative effects of
perfectionism, especially at midlevels of perfectionism (Blatt et
al., 1996). Thus, although differences among treatment condi-
tions appear to be minimal, significant outcome differences
emerged as a consequence of personality qualities patients bring
to the treatment process.
In terms of qualities of the therapist, Krupnick et al. (1996),
using the Vanderbilt Therapeutic Alliance Scale, found that
mean therapeutic alliance, assessed across the 3rd, 9th, and
15th sessions of the T D C R E was significantly related to out-
come across treatment groups. This effect was determined,
however, primarily by the patients' rather than the therapists'
contributions to the alliance (Krupnick et al., 1996). In sum-
mary, and consistent with prior reports (e.g., Blatt et al., 1996;
Burns & Nolen-Hoeksema, 1992; Horvath & Symonds, 1991;
Krupnick et al., 1996), findings suggest that therapeutic gain in
the TDCRP is significantly influenced by interpersonal dimen-
sions of the treatment process--by patient and therapist capac-
ity to establish a therapeutic relationship. Research has identi-
fied aspects of the patients' contribution to this relationship and
therapeutic outcome, but the therapist's contributions still re-
CM in the pharmacotherapy conditions included creating a sup-
portive interpersonal context for the treatment; psychological support
(e.g., conveying a sense of hope and optimism, providing reassurance);
instruction, education, and provision of information; simple advice
(e.g., encouraging greater physical activity); and ventilation of feelings.
Interventions not permitted included focusing on specificpsychological
themes, especially those related to interpersonal relations or cognitive
distortions; interpretations, especially those related to possible depress-
ogenic influences; clarification of the patient's feelings regarding sig-
nificant others or the therapist; specific behavioral instructions (apart
from simple advice); explorations of the psychodynamics of depression
(e.g., an examination of suppressed anger or shame); or encouraging
any "involved interpersonal interaction" (Fawcett et al., 1987, p. 321 ).
The intensity of initial training was comparable for pharmacotherapists
and psychotherapists (with videotaping of all sessions and weekly
supervision), and the intensity of monitoring during the actual study
was also equivalent ( i.e., supervisors continued to reviewvideotapes and
to provide monthly consultation).
1278 BLATT, SANISLOW, ZUROFF, AND PILKONIS
main relatively obscure. To explore more fully the therapists'
contributions, the present analyses sought to identify character-
istics o f more effective therapists by comparing therapists at
three levels of therapeutic efficacy, as defined by the average
therapeutic gain achieved by the patients of each therapist in
the TDCRP.
Method
Factor analysis of the residualized change scores of the five primary
outcome measures at termination in the TDCRP (HRSD, BDI, GAS,
SCL-90, and SAS) revealed that they loaded substantially (p > .79) on
a single factor with an eigenvalue of 3.78, accounting for 75.6% of the
variance (Blatt et at., 1996). No other eigenvalue approached 1.0, indi-
cating that this factor is a consistent measure of therapeutic change.
Thus, a composite of these 5 residualized change scores at termination
was used as the measure of therapeutic outcome for each patient} This
composite measure, averaged for all patients seen in active treatment by
each therapist, yielded an estimate of each therapist's overall therapeu-
tic et~cacy. The distribution of these mean outcome scores was divided
into thirds, thus identifying three groups of therapists: (a) more
effective, moderately effective, and less effective. As is discussed in the
Results section, several effects were nonlinear (i.e., the mean of the mid-
dle group of therapists did not lie midway between the means for the
other two groups). Thus, it seemed inappropriate to use the mean level
ofiherapists' ett~cacy as a continuous measure as well as statistical tech-
niques that assume linearity (e.g., correlations and linear regressions).
Differences among the therapists were clearer when three tiers of thera-
pists were compared.
Demographic characteristics and details of professional history were
available for the 28 therapists: (a) age, sex, race, marital status, religion,
and profession (MD or PhD level ); (b) years of general clinical experi-
ence and in treating seriously depressed patients, percentage of prior
depressed patients treated by psychotherapy alone, medication alone,
or a combination of psychotherapy and pharmacotherapy; and (c) kinds
of treatment ever used in treating seriously depressed patients (long-
term dynamic psychotherapy, short-term dynamic therapy, cognitive-
behavioral therapy, behavior therapy, eclectic psychotherapy, pharma-
cotherapy, and ECT). Therapists were also asked to rate their attitudes
and expectations, on a 7-point scale, about the etiology of major depres-
sion and components they considered essential to successful outpatient
treatment of major depression. 3 They also indicated the percentage of
seriously depressed outpatients whom (a) they had treated who had at
least moderate improvement; (b) they expected to show at least moder-
ate improvement with adequate treatment and without treatment; (c)
when change might be first manifested in treatment; and (d) the length
of time generally necessary for successful treatment of seriously de-
pressed outpatients.
Attitudes about the etiology of depression were grouped into four
clusters: biological (biochemical imbalance and genetic predis-
position), environmental events (adverse life experiences, stressful
events, chronic environmental stress), interpersonal (pathogenic social
relations or social isolation), and psychological factors (learned mal-
adaptive behavior, distorted cognitions, predisposing personality traits,
sense of helplessness).
Components considered important in the successful treatment of se-
riously depressed patients were also clustered into categories: medical
(pharmacotherapy), cognitive-behavioral (reinforcing adaptive behav-
ior, substituting realistic cognitions), interpersonal (developing more
adequate interpersonal behavior), psychodynamic (genetic reconstruc-
tion and uncovering unconscious conflict), and general therapeutic
(establishing a supportive therapeutic relationship, helping patients to
feel less helpless).
Characteristics of the three groups of therapists were compared using
one-way ANOVA and chi-square. In addition, the group of more
effective therapists was contrasted with the other two groups.
Results
Table 1 presents the distribution o f mean therapeutic out-
come scores of the 28 therapists based on the patients they saw
in active treatment. These scores ranged from 1.11 to - 1.02,
with a mean o f - 0 . 0 7 5 (SD = 0.49). Ten of the 28 therapists
with a mean therapeutic outcome score that ranged from 1.11
to 0.16 were labeled less effective ( G r o u p 1). Nine therapists
with a mean outcome score that ranged from 0.09 to - 0 . 2 7 were
labeled moderately effective ( G r o u p 2), and 9 therapists with a
mean outcome score that ranged from - 0 . 4 4 to - . 1.02 were
labeled as more effective ( G r o u p 3). N o significant difference
was found in therapist efficacy across the three active treatment
groups (CBT, IPT, and I M I - C M ) or across the three research
sites. The Treatment x Site interaction was also not significant.
Table 1 also presents some demographic features and aspects of
the usual clinical practice of these therapists.
To test for possible artifacts that might influence compari-
sons of these three groups o f therapists, the demographic (i.e.,
age, sex, education, marital status) and pretreatment clinical
characteristics (i.e., BDI, H R S D , GAS, SCL-90, SAS, and the
Need for Approval and Perfectionism factors of the DAS) of the
patients randomly assigned to these therapists were compared.
Results indicate that the patients treated by less effective, mod-
erately effective, and more effective therapists were essentially
equivalent at the start of treatment. In addition, the three
groups of therapists were assigned approximately the same av-
erage number o f patients in the three active treatment condi-
tions (6.20 [SD = 2.90], 6.67 [SD = 2.00], and 6.11 [SD =
2.57 ], from less to more effective, respectively). Four therapists
had less than 3 patients complete treatment; 3 of these thera-
pists were in the less effective group, two of them providing IMI-
C M and the other CBT. A fourth therapist from the more
effective group, who treated only 1 patient who completed treat-
ment in IPT, had relocated to another city and could not con-
tinue to participate in the research program. Thus, less effective
therapists tended to have fewer treatment completers (shown
later in Table 4).
2 Because each of the residualized outcome scores were calculated by
regressing the score of a particular scale at termination on the pretreat-
ment value of that variable, a negative score indicates more effective
outcome. The resulting factor scores are scaled in the same direction,
with negative scores indicating better outcome. In obtaining the mean
therapeutic outcome score for each therapist, we included only patients
in active treatment and did not include those patients who were in the
placebo group because they would not be expected to demonstrate as
much therapeutic change as patients in active treatment. Thus, the
mean therapeutic outcome score was calculated for each therapist based
only on patients in active treatment (CBT, IPT, and IMI-CM). Six of the
131 patients who completed one of the three forms of active treatments
lacked one or more of the five primary outcome measures. These pa-
tients were omitted from these analyses.
3 One of the 28 therapists, a male psychiatrist providing IPT who was
in the moderately effective group, failed to complete the form assessing
attitudes and expectations about the etiology and treatment of depres-
sion. Thus, the sample size in these comparisons were reduced to 10, 8,
and 9 for the less effective to the more effective therapists, respectively.
 CHARACTERISTICS OF EFFECTIVE THERAPISTS 1279

Table 1
Demographic and Practice Characteristics of Therapists in the NIMH TDCRP
Demographics % Clinical practice
Mean therapy No. (and %) of therapy
efficacy completersa Treatment Sex Profession Therapy alone Medication alone Combination

Group 1: Less effective

1.11 (less) 5 (83) IPT M MD 66 0 34
.91 7 (78) CBT M PhD 50 0 50
.49 7 (64) CBT M MD 25 0 75
.37 1 (50) IMI M MD 10 10 80
.31 2 (40) 1MI M MD 20 0 80
.29 3 (60) CBT F PhD 98 0 2
.29 5 (50) CBT M MD 2 75 23
.27 1 (25) CBT F PhD 85 0 15
.21 5 (83) IPT M MD 50 1 49
• 16 3 (75) IMI M MD 15 5 80

Group 2: Moderately effective

•09 4 (67) IMI M MD 50 0 50
.02 3 (100) IMI F MD 5 10 85
.00 4 (50) IPT M PhD 60 0 40
- . 13 6 (86) IPT M MD 40 0 60
- . 15 4 (100) IPT F MD 10 70 20
- . 15 4 (57) IPT M MD 75 0 25
- . 18 4 (50) IMI M MD 10 0 90
-.24 9 (100) IM1 M MD 10 40 50
-.27 5 (62) IMI M MD 5 10 85

Group 3: More effective

-.44 l (100) IPT F PhD 100 0 0
-.50 3 (60) CBT M PhD 85 0 15
-.54 7 (100) CBT M PhD 95 0 5
-.55 7 (78) IPT M PhD 75 0 25
-.56 7 (88) IPT F PhD 95 0 5
-.61 7 (88) CBT M PhD 85 0 15
-.63 4 (80) IPT M MD 99 0 I
-.67 4 (50) IMI F MD 20 40 40
- 1.02 (more) 3 (75) IMI F MD 10 20 70

Note. NIMH TDCRP = National Institute of Mental Health Treatment of Depression Collaborative Research Program; IPT = interpersonal
psychotherapy; M = male; CBT = cognitive-behavioral psychotherapy; IMI = imipramine; F = female.
a Number of patients completing active treatment (CBT, IPT or imipramine plus clinical management) whether or not they had all five primary
outcome measures.

To assess the stability o f the differences in therapeutic efficacy
among the three groups o f therapists, several ANOVAs were
conducted to evaluate whether the variability of therapeutic
outcome among the therapists was greater than the variability
o f therapeutic outcome within therapists. Patients in the PLA-
C M condition were not included in these analyses. Also, the 4
therapists who had less than 3 patients complete treatment were
not included; a total o f 24 therapists had sufficient data to be
included in these analyses. Therapist was treated as a fixed effect
that was nested within the three levels o f active treatment (CBT,
IPT, and I M I - C M ) , and patients were nested within therapists.
The error t e r m used in evaluating therapist effects was the aver-
age variability o f patients' outcome scores within therapists.
The overall therapist effect across the 24 therapists was not sig-
nificant, F ( 2 1 , 90) = 1.35, p < .17, suggesting that there were
no significant differences among the entire group o f 24 thera-
pists. This comparison, however, is unduly conservative because
it aggregates differences a m o n g the more effective and less
effective therapists, which might be substantial, as well as
differences among moderately effective, more effective, and less
effective therapists, which would be smaller. Accordingly, we
conducted a planned comparison of the outcomes o f the 73
completer patients who were treated by the 8 more effective and
the 7 less effective therapists who had at least 3 patients com-
plete treatment. The therapist effect was significant, F ( 12, 58)
= 1.99, p < .05, indicating that the patients o f the more effective
therapists improved to a significantly greater degree than the
patients of less effective therapists. We also compared the out-
comes of patients seen by the 8 more effective therapists with
the outcome o f the patients o f the 9 therapists in the moderately
effective group and also compared the outcome o f patients seen
by therapists in the moderately and less effective groups. In nei-
ther case was the effect of therapist significant. Thus, moder-
ately effective therapists were not significantly different from
their more effective and less effective colleagues, but the more
effective and less effective therapists differed significantly.
Interestingly, the mean variability o f the therapeutic outcome
scores within each therapist was significantly correlated with
1280 BLATT, SANISLOW, ZUROFF, AND PILKONIS

Table 2
Demographic and Professional Characteristics of Less Effective and More Effective Therapists
×2 or F
Therapeutic effectiveness
Contrast
Less Moderate More (Groups
Characteristic (n = 10) (n = 9) (n = 9) Overall l and 2 vs. 3)

Demographic ×2 (2) ×2 ( 1)

Sex (male/female) 8/2 7/2 5/4 1.65 1.64

Profession (MD/PhD) 7/3 8/1 3/6 6.27* 5.53*
Types of therapy used with
depressed patients (yes/no)
Long-term dynamic 7/3 7/2 6/3 0.29 0.15
Short-term dynamic 9/1 9/0 8/1 0.23 1.84
CBT 6/4 4/5 4/5 0.62 0.16
BT 4/6 2/7 3/6 0.70 0.01
Eclectic 6/4 8/1 5/4 2.73 0.92
Pharmacotherapy 9/1 8/1 4/5 6.61 * 6.60"*
ECT 5/5 7/2 1/8 8.12* 6.65**

M (and SD) ANOVA (F)

Clinical experience (in years) 11.0 (5.8) 10.9 (8.2) 12.0 (7.50) 0.07 0.36
Previous clinical experience
with depressed patients
Hr/week 16.4(8.7) 16.7(10.4) 13.1 (7.2) 0.45 0.95
Totalno. 129.5(93.1) 211.7(151.6) 185.0(215.8) 0.68 0.22
% Clinical practice devoted to:
Psychotherapy alone 42.1 (33.0) 29.4 (27.1) 73.8 (34.4) 4.69* 2.96**
Medication alone 9. l (23.4) 14.4 (24.6) 6.7 (14.1) 0.31 0.59
Combination 48.8 (29.5) 56.1 (26.1) 19.6 (22.9) 4.86* 3.08**

Note. CBT = cognitive-behavioral therapy; BT = behavioral therapy; ECT = electroconvulsivetherapy;

ANOVA = analysis of variance.
*p < .05. **p < .01.

their average level of therapeutic outcome, r(24) = .64; p <
.001. More effective therapists had significantly less variability
among the therapeutic outcome of their patients. The average
within-therapist variability for the 24 therapists with three or
more treatment completers in the less, moderate, and more
effective groups was 1.18, 0.81, and 0.59, respectively, a statisti-
cally significant (p = .023) difference. In summary, the distinc-
tion among the three groups of therapists, based on the average
therapeutic outcome of the patients they treated, appears to be
a reasonably reliable and accurate estimate of the level of the
therapeutic efficacy of the therapists.
As indicated in Table 2, the three groups of therapists did not
differ significantly in age, sex, race, religion, marital status, and
level of clinical experience. No significant differences were
found in their prior use of long- or short-term dynamic therapy,
CBT, behavior therapy, and eclectic psychotherapy. Significant
differences emerged, however, in their use of pharmacotherapy
and ECT. The ratio of use to nonuse of pharmacotherapy by
level of therapeutic efficacy, from less to moderately to more
effective was 9:1, 8:1, and 4:5, respectively (p = .037); for ECT,
the ratio of use to nonuse was 5:5, 7:2, and 1:8, respectively (p
= .017 ). These results are consistent with significant difference
in professional training of the therapists. The ratio of MD-level
to PhD-level therapists in each of the three levels of therapeutic
efficacy (from less to moderately to more effective) was 7:3, 8:
1, and 3:6, respectively. A significantly(p < .04) higher percent-
age of PhD-level than MD-level therapists (60.0%, as compared
with 16.7%, respectively) were in the therapeutically more
effective group.
Table 2 also presents a comparison of the general clinical
practice of these three groups of therapists: the percentage of
patients previously treated with psychotherapy alone, medica-
tion alone, or a combination of psychotherapy and pharmaco-
therapy. The more effective therapists reported that they had
treated significantly more of their depressed outpatients with
psychotherapy alone and relatively rarely used medication, ei-
ther alone or in combination with psychotherapy. Less and
moderately effective therapists reported that they more often
use medication, either alone or in combination with psycho-
therapy. In summary, these findings indicate that more effective
therapists had a psychological rather than a biological orienta-
tion in their treatment of depressed outpatients.4
4 These overall findings were statistically significant despite the fact
that the 3 most effectivetherapists in this study were psychiatrists and
that the two most effectivetherapists were female in the pharmacother-
apy condition (IM1-CM and PLA-CM) whose preferred mode of inter-
vention with depressed outpatients was a combination of psychotherapy
and medication (70% and 40%, respectively). Because the characteris-
tics of these two more effective therapists run counter to the predomi-
 CHARACTERISTICS OF EFFECTIVE THERAPISTS 1281

Table 3
Clinical Attitudes and Expectations of Less Effective and More Effective Therapists
ANOVA (F)
M (and SD) for therapeutic effectiveness
Contrast
Less Moderate More (Groups
Attitudes and expectations (n = 10) (n = 8) (n = 9) Overall 1 and 2 vs. 3)

Etiology of depression
Biological factors 4.2 (1.4) 4.5 (1.2) 4.0(1.2) 0.40 0.72
Traumatic events 4.4 (0.9) 3.6 (0.8) 4.5 (0.8) 2.88~ 1.48
Interpersonal difficulties 4.9 (1.1) 3.9 (1.6) 4.8(1.4) 1.48 0.71
Psychological factors 5.1 (0.8) 3.8 (0.7) 4.4(0.7) 7.21"* 0.06
Components of successful treatment
Medication 4.5 (0.8) 4.6 (1.3) 3.7(0.9) 2.37 2.17"
Cognitive-behavioral 5.2 (1.5) 4.6 (1.2) 4.8(0.9) 0.48 0.15
Interpersonal 4.9 (1.3) 5.12 (1.0) 5.0 (0.9) 0.09 0.03
Psychodynamic 2.8 (1.0) 3.1 (1.0) 2.9(1.3) • 0.20 0.18
General therapeutic 5.45 (1.4) 6.3 (0.8) 5.4(1.0) 1.58 0.93
% Depressed patients previously treated who improved 82.3 (12.9) 83.8 (11.6) 81.4(16.5) 0.06 0.28
% Depressed patients expected to show improvement
With treatment 83.9 (9.5) 85.0 (6.0) 81.4(8.5) 0.42 0.89
Without treatment 47.8 (14.8) 48.8 (10.9) 40.0(3.0) 1.05 1.45
Sessions necessary for depressed patients to begin to manifest
therapeutic change 3.4 (1.3) 3.6 (1.9) 5.3(3.0) 2.18 2.06*
Time necessary for successful treatment of depressed patients
(in months) 6.2 (2.2) 6.5 (5.0) 10.1 (10.3) 0.97 1.38

Note. ANOVA= analysis of variance.

*p<.05. **p<.01. ~tp=.076.

Another approach to identifying characteristics of more and
less effective therapists was to compare their attitudes and ex-
pectations about the etiology and treatment of seriously de-
pressed outpatients. As indicated in Table 3, the relative lack of
nant findings in these data, we explored the possibility that these two
therapists were unusually effective in establishing a therapeutic alliance
with their patients. Although no significant differences were found
among the three groups of more, moderate, and less effective therapists
in the mean or standard deviation of their BL-RI scores after the second
treatment hour (see Table 4), it is noteworthy that the standard devia-
tions of the BL-RI scores of these two therapists in IMI-CM were 23.10
and 21.88, respectively, substantiallybelow the mean standard deviation
of 35.34 for the other 26 therapists. These two therapists were also
among the most effective of the 10 therapists in PLA-CM. The mean
therapeutic outcome score of the 10 therapists treating patients in PLA-
CM condition ranged from 1.01 to -0.88, roughly approximating the
range of therapeutic efficacy scores of the 28 therapists in the three ac-
tive treatment conditions that ranged from 1.11 to - 1.02. These 2 ther-
apists ranked third and fourth highest in therapcutic efficacy among the
10 therapists treating patients in the PLA-CM condition.
The therapist with the second best therapeutic efficacy score, however,
had a low percentage of patients who completed treatment--50% (4 of
8) in IMI-CM (as compared with an average of 72% for the other 27
therapists in the three active treatment conditions) and only 44% (4 of
9) of her patients in PLA-CM completed treatment (as compared with
an average of 66% for the other 9 therapists). In contrast, the most
effective therapist had 75% (3 of 4) treatment completers in IMI-CM
and 100% (3 of 3) in PLA-CM. It is noteworthy that this therapist's high
level of therapeutic effectiveness was accomplished while seeing patients
for a relatively brief time each week (approximately 25 min) as part of
CM in IMI and PLA--a procedure designed as a minimal therapeutic
condition to provide only general support and encouragement.
emphasis placed on medication significantly differentiated
more effective therapists from moderately and less effective ther-
apists. Attitudes about etiology of depression also related sig-
nificantly to therapeutic efficacy. Less and more effective thera-
pists tended to consider psychological factors and adverse envi-
ronmental experiences as more central to the etiology of
depression than moderately effective therapists.
No significant differences were found between therapeutic
efficacy and optimism about treating seriously depressed outpa-
tients. All therapists expected over 80% of seriously depressed
patients to show moderate improvement with treatment. Also
more effective, as compared with less effective, therapists ex-
pected therapy to require somewhat more sessions before de-
pressed patients manifested treatment-related changes and that
successful outpatient treatment of serious depression required a
longer time. This latter finding, however, did not reach statistical
significance.
Table 4 presents additional data about the three groups of
therapists--the percentage of patients completing treatment
and mean and standard deviation of the B-L RI scores they re-
ceived from their patients after the second treatment hour. Al-
though more effective therapists had a greater percentage of pa-
tients complete treatment and greater consistency (lower stan-
dard deviation) in their B-L RI scores than did moderate or less
effective therapists, these differences were not significant.
Discusfion
The present analyses of the data of the T D C R P indicate that
significant differences exist in therapeutic efficacy among thera-
1282 BLATT, SANISLOW, ZUROFE AND PILKONIS
Table 4
Treatment Characteristics of Less Effective and More Effective Therapists
Within the NIMH TDCRP
M (and SD) for therapeutic effectiveness
Less
Clinical characteristics (n = 10)
% Patients seen who completed
treatment 60.8 (19.6)
Barrett-Lennard Relationship
Inventory
M 61.6 (22.5)
SD 36.6(20.4)
Note. NIMH TDCRP = National Institute of Mental Health Treatment of Depression Collaborative Re-
search Program; ANOVA= analysis of variance.
]"p= .108.
pists, even within the experienced and well-trained therapists in
the TDCRP. Differences in therapeutic efficacy were indepen-
dent of the type of treatment provided or the research site and
not related to the therapists' level of general clinical experience
or in treating depressed patients. Differences in therapeutic
efficacy, however, were associated with basic clinical orienta-
tion, especially about treatment. More effective therapists had
a more psychological rather than biological orientation to the
clinical process. They reported using predominantly psycho-
therapy with depressed patients and rarely using biological
interventions (i.e., medication and ECT). The more effec-
tive therapists were more likely to be psychologists than psychi-
atrists. (The same basic findings were obtained when the thera-
peutic outcome of patients in the PLA-CM condition was in-
cluded in the estimates of therapeutic efficacy of therapists in
the medication condition.)
Less effective therapists, somewhat like the more effective
therapists, reported that they primarily tend to use psychother-
apy in their clinical practice (42.1% of the time) but, more often
than the more effective therapists, they combine their psycho-
therapeutic efforts with medication with almost half ( 48.6% ) of
the depressed patients they treat. More effective therapists, in
contrast, primarily use psychotherapy alone (73.8%) and only
occasionally (19.6%) combine their psychotherapy with medi-
cation. Moderately effective therapists, as compared with less
and more effective therapists, primarily use medication, either
alone (14.4%) or in combination with psychotherapy (56.1% ),
and relatively rarely use psychotherapy alone (29.4%). Thus,
moderately effective therapists appear to be more biologically
oriented. Less effective therapists, like the more effective thera-
pists, are primarily interested in psychotherapy but combine
their psychotherapy with the use of medication more often than
effective therapists. Additionally, more effective therapists, com-
pared with less and moderately effective therapists, expect ther-
apy to require more treatment sessions before patients begin to
manifest therapeutic change.
It is noteworthy that most of the significant differences be-
tween relatively more and less effective therapists were found in
reports about their clinical practice with depressed outpatients.
Relatively few significant findings were obtained when compar-
ANOVA F(2, 25)
Contrast
Moderate More (Groups
(n = 9) (n = 9) Overall 1 and 2 vs. 3)
74.7 (21.8) 79.8 (16.7) 2.43t 1.52
58.7 (15.3) 67.3 (l 1.2) 0.57 1.02
30.4 (13.8) 25.0 (10.6) 1.19 1.26
ing attitudes about the etiology of depression or about tech-
niques considered essential to successful treatment. Although
attitudes of therapists about the etiology of depression and the
nature of the therapeutic process may not have a direct relation-
ship to therapeutic efficacy, these attitudes may still influence
therapeutic outcome, possibly indirectly in interaction with the
attitudes and expectations that their patients have about the eti-
ology and treatment of depression. We will examine this possi-
bility in subsequent analyses of the TDCRP data.
The findings of the present analyses also raise an important
question about the nature of mutative forces in the outpatient
treatment of depression. The overall results indicate that quali-
ties of the therapist are important dimensions that appear to
influence therapeutic outcome. In addition, as noted in Foot-
note 4, the fact that one therapist could be so very effective even
when providing only CM with patients receiving a placebo, sug-
gests that the therapeutic alliance established by a talented cli-
nician may be an essential component of effective treatment.
These conclusions are consistent with prior findings (e.g., Blatt
et al., 1996; Burns & Nolen-Hoeksema, 1992; Horvath & Sy-
monds, 1991; Krupnick et al., 1996) that therapeutic outcome
is significantly influenced by interpersonal dimensions of the
treatment process--by personal qualities patient and therapist
bring to the treatment process and their ability to establish an
effective therapeutic relationship. Future research should be di-
rected toward exploring these interpersonal dimensions in de-
tail, such as evaluating the therapeutic sessions of more effective
therapists. As typescripts of the TDCRP treatment sessions be-
come available, for example, aspects of the therapeutic relation-
ship and therapeutic technique of the more effective therapists
in the TDCRP should be studied more fully, especially the fe-
male psychiatrist in the pharmacotherapy condition of the
TDCRP who, as noted in Footnote 4, had the highest therapeu-
tic efficacy score, as well as a very high percentage (86%) of
treatment completers in both IMI-CM and PLA-CM.
In summary, significant differences were found in the
TDCRP among therapists who achieved different levels of ther-
apeutic efficacy. These differences are impressive because they
occurred in a relatively homogenous group of well-trained and
experienced therapists across three well-specified treatment
 CHARACTERISTICS OF EFFECTIVE THERAPISTS
conditions in three independent research sites. These findings
support the contention that it is important to differentiate
among therapists and to include dimensions o f therapists in
studies o f therapeutic outcome.
References
Barrett-Lennard, G. T. (1962). Dimensions of therapist responses as
causal factors in therapeutic change. Psychological Monographs,
76(43, Whole No. 562).
Barrett-Lennard, G. T. ( 1985 ). The Relationship Inventory now: Issues
and advances in theory, method, and use. In L. S. Greenberg & W. M.
Pinsof (Eds.), The psychoanalytic process: A research handbook ( pp.
439-476). New York: Guilford.
Beutler, L. E., Crago, M., & Arizmendi, T. G. (1986). Therapist vari-
ables in psychotherapy process and outcome. In S. L. Garfield & A. E.
Bergin (Eds.), Handbook of psychotherapy and behavior change ( 3rd
ed., pp. 257-310). New York: Wiley.
Beutler, L. E., Machado, P., & Neufeldt, S. (1994). Therapist's vari-
ables. In A. Bergin & S. Garfield (Eds.), Handbook of psychotherapy
and behavior change (pp. 229-269 ). New York: Wiley.
Blatt, S. J., Quinlan, D. M., Pilkonis, E A., & Shea, T. (1995). Impact
of perfectionism and need for approval on the brief treatment of de-
pression. Journal of Consulting and Clinical Psychology, 63, 125-
132.
Blatt, S. J., Zuroff, D. C., Quinlan, D. M., & Pilkonis, P. (1996). Inter-
personal factors in brief treatment of depression: Further analyses of
the NIMH Treatment of Depression Collaborative Research Pro-
gram. Journal of Consulting and Clinical Psychology, 64, 162-17 I.
Bums, D. D., & Nolen-Hoeksema, S. (1992). Therapeutic empathy
and recovery from depression in cognitive-behavioral therapy: A
structural equation model. Journal of Consulting and Clinical Psy-
chology, 60, 441-449.
Crits-Christoph, E (1992). The efficacy of brief dynamic psychother-
apy: A meta-analysis. The American Journal of Psychiatry, 149, 151-
158.
Crits-Christoph, P., Baranackie, K., Kurcias, J., & Beck, A. T. ( 1991 ).
Meta-analysis of therapist effect in psychotherapy outcome studies.
Psychotherapy Research, 1, 81-91.
Crits-Christoph, P., & Mintz, J. (1991). Implications of therapists
effects for the design and analysis of comparative studies of psycho-
therapies. Journal of Consulting and Clinical Psychology, 45, 1001-
1004.
Elkin, I. (1994). The NIMH Treatment of Depression Collaborative
Research Program: Where we began and where we are now. In A. E.
Bergin & S. L. Garfield (Eds.), Handbook of psychotherapy and be-
havior change (4th ed., pp. 114-135 ). New York: Wiley.
Elkin, I., Shea, M. T., Watkins, J. T., Imber, S. D., Sotsky, S. M., Collins,
J. E, Glass, D. R., Pilkonis, P. A., Leber, W. R., Dockerty, J. P., Fies-
ter, S. J., & Parloff, M. B. (1989). NIMH Treatment of Depression
Collaborative Research Program: General effectiveness of treat-
ments. Archives of General Psychiatry, 46, 971-983.
Fawcett, J., Epstein, P., Fiester, S. J., Elkin, I., & Autry, J. H. (1987).
Clinical management-imipramine/placebo administration manual:
NIMH Treatment of Depression Collaborative Research Program.
Psychopharmacology Bulletin, 23, 309-324.
Frank, A. D. (1959). The dynamics of the psychotherapeutic relation-
ship: Determinants and effects of the therapist's influence. Psychia-
try, 22, 17-39.
Gurman, A.S., & Razin, A.M. (1977). Effective psychotherapy: A
handbook of research. New York: Pergamon Press.
Hamilton, M. A. (1960). A rating scale for depression. Journal of Neu-
rology, Neurosurgery, and Psychiatry, 6, 56-62.
Hamilton, M. A. (1967). Development of a rating scale for primary
1283
depressive illness. British Journal of Social and Clinical Psychology,
6, 278-296.
Holloway, E. L., & Wampold, B. E. (1986). Relation between concep-
tual level and counseling-related tasks: A recta-analysis. Journal of
Counseling Psychology,, 33, 31O-319.
Horvath, A. O., & Symonds, B. D. ( 1991 ). Relation between working
alliance and outcome in psychotherapy: A meta-analysis. Journal of
Counseling Psychology, 38, 139-149.
Krupnick, J. L., Sotsky, S. M., Simmens, S., Moyer, J., Elkin, I., Wat-
kins, J., & Piikonis, P. A. (1996). The role of the therapeutic alliance
in psychotherapy and pharmacotherapy outcome: Findings in the
National Institute of Mental Health Treatment of Depression Collab-
orative Research Program. Journal of Consulting and Clinical Psy-
chology, 64, 532-539.
Lafferty, P., Beutler, L. E., & Crago, M. (1989). Differences between
more and less effective psychotherapies: A study of select therapist
variables. Journal of Consulting and Clinical Psychology, 57, 76-80.
Lambert, M. J. ( 1989 ). The individual therapist's contribution to psy-
chotherapy process and outcome. Clinical Psychology Review,, 9, 469-
485.
Luborsky, L., & Auerbach, A. H. ( 1985 ). The therapeutic relationship
between working alliance and outcome in psychotherapy: The re-
search evidence and its meaning for practice. In R. E Hales & A. J.
France (Eds.), Annual review (Vol. 4, pp. 550-561 ). Washington,
DC: American Psychiatric Press.
Luborsky, L., Crits-Christoph, P., McLellan, A. T., Woody, G., Piper,
W., Liberman, B., Imber, S., & Pilkonis, P. (1986). Do therapists vary
much in their success? Findings from four outcome studies. American
Journal of Orthopsychiatry, 56, 501-512.
Luborsky, L., Diguer, L.A., McLellan, A.T., & Woody, G. (1995,
July). The psychotherapist as a neglected variable: New studies of
each therapist's benefits to their patients. Paper presented at the meet-
ings of the Society for Psychotherapy Research, Vancouver, British
Columbia, Canada.
Luborsky, L., McLellan, A. T., Woody, G. E., O'Brien, C. P., & Auer-
bach, A. ( 1985 ). Therapist success and its determinants. Archives of
General Psychiatry, 42, 602-611.
Luborsky, L., Woody, G. E., McLellan, A. T., O'Brien, C. P., & Rosenz-
weig, J. (1982). Can independent judges recognize different psycho-
therapies? An experience with manual-guided therapies. Journal of
Consulting and Clinical Psychology, 50, 49-62.
Lyons, L. C., & Woods, P. J. ( 1991 ). The efficacy of rational-emotive
therapy: A quantitative review of the outcome research. Clinical Psy-
chology Review.. 11, 357-370.
McLellan, T., Woody, G. E., Luborsky, L., & Goehl, L. (1988). Is the
counselor an "active ingredient" in substance abuse rehabilitation?
An examination of treatment success among four counselors. The
Journal of Nervous and Mental Disease, 176, 423-430.
Miller, R. C., & Berman, J. S. ( 1983 ). The efficacy of cognitive behavior
therapies: A quantitative review of the research evidence. Psychologi-
cal Bulletin. 94, 39-53.
Najavits, L. M., & Strupp, H. H. (1994). Differences in the effective-
ness of psychodynamic therapists: A process-outcome study. Psycho-
therapy, 31, 114-123.
Odinsky, D. E., & Howard, K. (1980). Gender and psychotherapeutic
outcome. In A. M. Brodsky & R. T. Hare-Mustin (Eds.), Women and
psychotherapy (pp. 3-34 ). New York: Guilford.
Parloff, M. B., Waskow, I. E., & Wolfe, B. E. (1978). Research on ther-
apist variables in relation to process and outcome. In S. L. Garfield
& A.E. Bergin (Eds.), Handbook of psychotherapy and behavior
changes (2nd ed., pp. 233-282). New York: Wiley.
Ricks, D. (1974). Supershrink: Methods of a therapist judged success-
ful on the basis of adult outcome of adolescent patients. In D. Ricks,
1284 BLATT, SANISLOW, ZUROFF, AND PILKONIS
M. Roff, & A. Thomas (Eds.), Life history research in psychopathol-
ogy (Vol. Ill ). Minneapolis: University of Minnesota Press.
Smith, M. L., Glass, G. V., & Miller, T. I. (1980). The benefits of psy-
chotherapy. Baltimore: John Hopkins University Press.
Sotsky, S. M., Glass, D. R., Shea, M. T., Pilkonis, E A., Collins, J. E,
Elkin, I., Watkins, J. T., Imber, S. D., Leber, W. R., Moyer, J., & Oli-
veri, M. E. ( 1991 ). Patient predictors of response to psychotherapy
and pharmacotherapy: Findings in the NIMH Treatment of Depres-
sion Collaborative Research Program. American Journal of Psychia-
try, 148, 997-1008.
Spitzer, R. L., Endicott, J., & Robins, E. (1978). Research diagnostic
criteria: Rationale and reliability. Archives of General Psychiatry, 35,
773-782.
Stein, D. M., & Lambert, M. J. (1984). On the relationship between
therapist experience and psychotherapy outcome. Clinical Psychol-
ogyReview, 4, 127-142.
Stiles, W. B., Shapiro, D. A., & Elliot, R. (1986). Are all psycho-thera-
pies equivalent?American Psychologist, 41, 161-180.
Svartberg, M., & Stiles, T. C. ( 1991 ). Comparative effects of short-term
psychodynamic psychotherapy: A meta-analysis. Journal of Consult-
ing and Clinical Psychology, 59, 704-714.
Weissman, A. N., & Beck, A. T. ( 1978, August-September). Develop-
ment and validation of the Dysfunctional Attitude Scale. A prelimi-
nary investigation. Paper presented at the Annual Convention of the
American Psychological Association, Toronto, Ontario, Canada.
Weissman, M. M., & Paykel, E. S. (1974). The depressed women: Study
of social relationships. Chicago: University of Chicago Press.
Received October 17, 1995
Revision received March 19, 1996
Accepted March 21, 1996 •